[SOLVED] Shell Scripting: gawk Record Separator and Regex

newmanium2001 · 03-06-2014, 11:40 AM

Hi,

I've been going crazy trying to figure out what I'm doing wrong with gawk here. Would appreciate any feedback I could get on this.

Given an input file like this:

Code:

*******************************************
*******************************************
foo bar
something something
*.somewhere.com
foo again

*******************************************
*******************************************

foo bar
*.somewhereelse.com
some more stuff

*******************************************
*******************************************

I want to read it in with gawk and use the two lines of asterisks as the record separator (RS). Here's what I've got so far, but it only creates one record in awk every time I try it:

Code:

gawk -W dump-variables=awk.out 'BEGIN {RS="^\*+"}  {
        print $0
        print NR
        print "\n"

}'

Output:

Code:

*******************************************
*******************************************
foo bar
something something
*.somewhere.com
foo again

*******************************************
*******************************************

foo bar
*.somewhereelse.com
some more stuff

*******************************************
*******************************************

1

What am I doing wrong with the RS regular expression?

grail · 03-06-2014, 11:57 AM

If you are going to use both lines of asterisk as the separator then you need to tell awk that. Also, I think your anchor is not helping

Something like this should work:

Code:

awk '{print "|"$0"|"}END{print NR}' RS='[*]+\n[*]+\n' file

I often find escaping something like an asterisk can also cause problems, so I prefer '[*]' instead.

newmanium2001 · 03-06-2014, 03:20 PM

That did it. Thanks! I was trying it simple by doing just one line of asterisks at a time, but yours works for both lines. And yeah, I'd agree that something funky is going on in trying to escape the asterisk in my example. Bracketing works much better!